🎵 Independent Musicians vs. Udio & Suno: When AI Music Becomes a Class Action 🤖
Generative AI music went from party trick to litigation target in about 18 months.
On October 15–16, 2025, a group of independent musicians filed back-to-back class actions against Udio (Uncharted Labs) and Suno, alleging that both platforms built their models by quietly ingesting thousands of copyrighted sound recordings and songs without permission. (Copyright Alliance)
These are not the big-label lawsuits you saw from the RIAA in 2024. These are artist-led cases, framed as a direct attack on independent musicians’ livelihoods, and they’re designed to sit alongside (not instead of) label litigation. (RIAA)
If you write, license, or build tools around music, this is the test case for what “AI infringement” actually looks like.
What Happened in the Udio and Suno Cases 🎤
On October 15, 2025, a group of musicians including David Woulard, Stan and James Burjek, Berk Ergoz, Hamza Jilani, Maatkara Wilson, Arjun Singh, Magnus Fiennes, and Michael Mell filed a class action in federal court against Uncharted Labs, Inc. (d/b/a Udio).
On October 16, 2025, the same group filed a substantially similar class action against Suno, Inc.
At a high level, they allege that each company:
- copied and ingested their sound recordings and musical works without permission,
- used that library to train text-to-music models, and
- now offers a commercial product that can spit out tracks resembling their music and style, undermining their licensing revenue.
A helpful way to see the structure:
| 🎧 Case | 🏛️ Court & Date | 🎼 Who’s Suing | 🎯 Core Allegations |
|---|---|---|---|
| Musicians v. Udio (Uncharted Labs) | U.S. federal court (filed Oct 15, 2025) | Group of independent musicians and composers | Udio allegedly scraped lyrics sites and stream-ripped tracks from platforms like YouTube to assemble a training corpus of copyrighted songs, without licenses or consent. |
| Musicians v. Suno | U.S. federal court (filed Oct 16, 2025) | Same musician plaintiffs | Suno allegedly engaged in similar ingestion of sound recordings and lyrics, then offered a product that can generate music strongly resembling plaintiffs’ works, sometimes when prompted with era/genre/artist-style cues. |
The complaints frame this as an attack on “the most vulnerable and valuable creators” in the music ecosystem, not just a technical copyright issue. (Digital Music News)
How the Lawsuits Say Udio and Suno Work Under the Hood 🧠🎚️
The pleadings and AI-litigation commentary describe a pipeline that looks roughly like this:
- Acquire recordings and lyrics at scale
- Udio is accused of scraping lyrics websites and stream-ripping songs from YouTube to build a training set of full songs and associated textual metadata. (Copyright Alliance)
- Suno is described in similar terms, with emphasis on ingestion of existing recordings whose rights belong to labels and artists. (Loevy + Loevy)
- Train text-to-music models
- The models are trained so that a text prompt (genre, era, mood, “in the style of X”) produces full-length audio tracks.
- Show ability to mimic existing songs
- Plaintiffs (and their experts) reportedly used targeted prompts referring to decade, genre, and artist descriptors; they claim Udio’s and Suno’s outputs closely resembled specific copyrighted recordings, sometimes enough to recognize hooks, chord progressions, or production signatures. (mckoolsmith.com)
- Monetize access
- Both platforms offer free and paid tiers for generating and downloading music; the complaints emphasize that this is not research but revenue-driven exploitation of artists’ catalogs. (Loevy + Loevy)
Udio’s viral “BBL Drizzy” track and early charting AI songs are background color: they demonstrate that AI-generated tracks can obtain millions of streams and chart placement, which makes the “market harm” argument much more concrete. (Wikipedia)
The Legal Theories Musicians Are Using ⚖️
These lawsuits are not just “you copied our vibe.” They stack several doctrines:
| ⚖️ Claim | 💿 What Plaintiffs Allege | 📌 Why It’s Dangerous for AI Music Platforms |
|---|---|---|
| Direct copyright infringement | Full songs and lyrics were copied into training datasets and stored/processed without license. (Copyright Alliance) | Focuses on the act of ingestion and storage of recordings, not just outputs. That’s the same weak point that drove Anthropic’s and label lawsuits. (RIAA) |
| Contributory / vicarious infringement | Platforms enable and profit from users generating infringing sound-alikes, and can control the models and filters. (mckoolsmith.com) | Ties liability to business decisions about model design and guardrails, not just training. |
| DMCA & “stream-ripping”–type violations | Udio and Suno allegedly bypassed technical measures by ripping songs from streaming platforms and scraping lyric sites. (Copyright Alliance) | Raises anti-circumvention and “you didn’t just scrape; you bypassed protections” narratives. |
| Right of publicity / privacy (e.g., Illinois) | Some claims allege unauthorized use of artists’ voices and personas, including under the Illinois Right of Publicity Act. (Copyright Alliance) | Moves the fight beyond copyright; even licensed compositions/masters might not solve persona-based claims. |
| Unjust enrichment & state UDAP claims | Platforms allegedly saved massive licensing costs and undercut the market for human commissions and licenses. (Copyright Alliance) | Sets the stage for disgorgement theories tied to platform valuation and revenue. |
Unsurprisingly, Suno is already trying to narrow artist suits (including related songwriter cases) by invoking fair use and arguing that training on recordings is transformative and socially beneficial. (Music Business Worldwide)
But as with the book cases, the battleground isn’t only fair use. It’s also about:
- whether mass ingestion of complete works crosses qualitative lines, and
- whether specific outputs cross into recognizable derivation.
How This Interacts with the Major-Label Cases 🎙️🏢
These artist-led class actions sit alongside the major-label lawsuits brought in 2024 by UMG, Sony, and Warner through the RIAA:
- In June 2024, labels sued Suno and Udio, calling them “nothing more than a stealing machine” that built models on vast troves of copyrighted recordings. (RIAA)
- By October 2025, UMG had settled with Udio, limiting how the platform can operate and reportedly restricting unlicensed models. (Pitchfork)
- By late November 2025, Warner had reached settlements with both Udio and Suno and pivoted to licensing deals, letting artists opt in to have their voices and compositions used in licensed AI models. (The Guardian)
So you end up with a two-track system:
- Labels securing corporate-level settlements and licensing frameworks;
- Artists continuing class actions focusing on individual rights and alleged harm to independent musicians’ income and control. (Loevy + Loevy)
That dual structure is important. Even if a platform cuts a deal with the majors, artist-level litigation can still continue, especially on right-of-publicity and individual copyright theories.
Why This Matters If You Make or Use AI Music 🧩
From a practical business perspective, these cases highlight three very immediate risks:
For AI music platforms
- Training on unlicensed recordings and lyrics is now a known litigation vector, not a hypothetical.
- “We relied on fair use” is being tested in court, but plaintiffs are also attacking the means of acquisition (stream-ripping, scraping behind interfaces) and the closeness of outputs to existing songs. (Copyright Alliance)
- Even if you settle with labels, you may still be facing artist and songwriter classes, as Suno and Udio now are. (Loevy + Loevy)
For labels, publishers, and production companies
- You need contracts that expressly address AI training, voice/likeness rights, and derivative use.
- You also need an internal policy about when you will license catalog for AI and how artists are compensated or allowed to opt out.
For musicians and composers
- These cases show that independent artists can organize and bring class actions, not just rely on labels.
- They also underscore the importance of copyright registration and clear ownership of masters and compositions before you start sending demand letters or joining a class. (Ari’s Take)
Contract Fallout: What Agreements Need to Say Now 🎛️📜
On the contracting side, there are at least three buckets that suddenly matter a lot more.
| 🧾 Agreement Type | 🧠 Key AI-Related Clauses to Revisit | 🎯 Litigation Lesson |
|---|---|---|
| Artist/label & publishing deals | Explicit grant (or denial) of rights to use masters and compositions for AI training, including text-to-music models; clear revenue participation or opt-out mechanisms. | Without explicit AI language, labels and platforms face arguments that catalog licensing never contemplated this use. |
| Sync & commissioned-music contracts | Clauses specifying whether the commissioner can feed delivered tracks into AI tools (e.g., to generate variants, stems, remixes) and whether the artist’s voice/timbre may be modeled. | Prevents later disputes when clients run tracks through AI to generate “sound-alikes” for new campaigns. |
| Platform Terms of Use & License Agreements | Representations about training data sources, DMCA/anti-circumvention compliance, and the extent to which user prompts or outputs are reused for training. | Udio/Suno complaints make clear that vague ToS plus aggressive ingestion is a litigation magnet. (Copyright Alliance) |
Without these, everyone downstream ends up guessing who actually owns what rights in the AI context.
Demand Letters Before and After Filing ✉️🎼
Artist-side law firms are already using pre-suit demand letters and sign-up forms to recruit musicians into these cases. (Loevy + Loevy)
On the musician side, a good pre-litigation letter typically:
- identifies specific recordings and compositions you own;
- asks whether and how they were used in training;
- reserves rights for both copyright and right-of-publicity claims;
- may demand removal from training sets and negotiation of licensing terms.
On the platform side, receiving such a letter now has to trigger:
- legal hold on training data, logs, and internal testing results;
- a careful review of how recordings were actually obtained (licensed catalog vs. YouTube rips vs. “research” sets);
- a decision whether to fight on fair use grounds, narrow usage, or shift toward licensed models like the ones Suno and Udio are now moving to with Warner/UMG. (The Guardian)
Treating these as routine customer complaints is exactly how you end up with a class action filed a week later.
The Bigger Picture: Udio and Suno as Templates, Not Outliers 🌐
Udio and Suno are not the only AI music platforms in the crosshairs; they’re just early, visible examples.
Between:
- the label suits pushing toward catalog-licensing deals, and
- the artist-led class actions attacking unlicensed training and sound-alike outputs,
we’re watching the contours of “responsible AI music” being litigated into existence.
The practical message:
- If you build AI music tools, you need a coherent story about where your training data came from and what contracts cover it.
- If you license music or represent artists, your agreements need explicit AI and training clauses, not generic copyright language.
- If you’re a musician, you now have a real, concrete roadmap—from demand letters to class actions—for pushing back when your catalog becomes someone’s training set without your consent.